High levels of CSF HIV-1 RNA, and associated neurobehavioral deficits, have been reported without commensurate serum viral levels (Ellis et al., 2000; Ellis et al., 1997; McArthur et al., 1997; Staprans et al., 1999; Wong et al., 1997). The implication is that serial lumbar puncture, although invasive, may be the best available method for monitoring the progression of HIV-related neurodegenerative processes. Sensitive noninvasive methods for detecting and monitoring HIV-related CNS effects are needed. The aim of the proposed studies is to evaluate the application of structural (SMRI) and functional (FMRI) imaging methods for predicting CNS viral replication. The approach is guided by previous studies suggesting that neurodegeneration in the striatum and diffuse cerebral white matter damage are the two most consistent structural abnormalities present in this population (Jernigan et al., 1993; Stout et al., 1998) and psychomotor slowing is the most prominent functional deficit. Since CSF HIV-1 RNA is currently the best available index of CNS viral replication, this measure will be the primary dependent variable for the proposed studies. Groups of neurobehaviorally impaired and unimpaired seropositive individuals, and yoked seronegative controls, will be studied with repeated sMRI/fMRI examinations at baseline, and 6 and 12 weeks after initiation (in the seropositive subjects) of aggressive HAART therapy aimed at reducing CSF HIV-1 RNA levels. Measures of white matter signal change, of caudate nucleus volume, and of task-related change in blood oxygenation in the caudate nucleus (during motor tasks) will be correlated with levels of CSF HIV-1 RNA within the seropositive subjects. The utility of these SMRI and FMRI measures in predicting CSF viral levels will be compared with that of behavioral measures, and possibly with that of other noninvasive methods (such as MR spectroscopy), in univariate and multivariate prediction models.